> On Nov. 18, 2013, 6:13 p.m., Ben Mahler wrote:
> > include/mesos/mesos.proto, lines 126-134
> > <https://reviews.apache.org/r/15542/diff/3/?file=387242#file387242line126>
> >
> >     Thanks for clarifying!
> >     
> >     Rather than offer two ways of doing it, let's mark the old one as 
> > deprecated so that we can eventually have frameworks only using this 
> > improved version. Or is there a reason to keep both versions indefinitely?
> >     
> >     How about we clean up the naming a little bit? ExecV seems to suggest 
> > too much about the implementation, what about s/ExecV/Command/ and 
> > s/args/arguments/ (we tend to avoid abbreviations):
> >     
> >     message Command {
> >       required string command = 1;
> >       repeated string arguments = 2;
> >     }
> >     
> >     ...
> >     required string value = 3; // Deprecated, please use command instead.
> >     optional Command command = 4;

I would be happy to change the names as you suggest.

Deprecating `value` makes a lot of sense, I will add the comment.

In the future -- maybe in a future update for this review? -- we could add an 
optional `bytes body` to the Command structure, which would allow users to pass 
a script -- shell or Python or Lua or PL/SQL -- which the slave could place in 
a temporary location and mark executable. This would be a good option for 
frameworks and tools that presently rely on shell control flow and glob 
expansion. (As does the Hadoop framework, for example.) The name of the file on 
disk where the `body` is placed would be randomly generated; the `command`, if 
present, could serve as the name of the command as it would show up in `ps`. 
(Rewriting $0 is something Postfix, Postgres and many other tools do to make it 
easier to interpret a process listing.) What do you think of that idea?


- Jason


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15542/#review29054
-----------------------------------------------------------


On Nov. 18, 2013, 1:31 a.m., Jason Dusek wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15542/
> -----------------------------------------------------------
> 
> (Updated Nov. 18, 2013, 1:31 a.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Ben Mahler, and Niklas Nielsen.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> Offer an execvp like interface for running tasks.
> 
> Review: https://reviews.apache.org/r/15542
> 
> 
> Diffs
> -----
> 
>   include/mesos/mesos.proto 655f86757487ddbe551fdcf53eb793e773ecdd34 
>   src/examples/python/test_framework.py 
> deca48e779ae099424fa73bb9a8ac5c419c5faf1 
>   src/launcher/executor.cpp b73ab479500a7347a38ba53acecfab9229f1080d 
>   src/launcher/launcher.cpp d5ab66704429a95eeb8eda5188e33d8e691221af 
>   src/launcher/main.cpp de64609905ee63096c0173fe7e64a1eafea5d6bf 
>   src/slave/process_isolator.cpp a6e9ed6a654972e8a51a9a033052e02ce44fe3e4 
> 
> Diff: https://reviews.apache.org/r/15542/diff/
> 
> 
> Testing
> -------
> 
> Ran Python test executor and `make check`.
> 
> 
> Thanks,
> 
> Jason Dusek
> 
>

Reply via email to